The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including ledger accounts, journal entries, and invoices, which creates a surface for indirect instructions to be processed by the agent.
Ingestion points: The skill retrieves data from ledger_accounts, journal_entries, and invoices via tool calls defined in SKILL.md.
Boundary markers: There are no explicit boundary markers or instructions to ignore embedded instructions within the retrieved financial data.
Capability inventory: The skill uses tool calls like well_get_schema to retrieve and aggregate financial data in SKILL.md.
Sanitization: No sanitization or validation of the retrieved financial data strings is described before the data is processed by the agent.

compte-de-resultat